MOTOR TOTAL LOSS

Every total loss settles on a residual value.
Nothing tells you whether it was a good one.

Serious people have tried to predict this figure for years, and the difficulty is real: the highest bid a damaged vehicle draws depends on which buyers happened to see it, in which week, in which country. ECX predicts that maximum, publishes how accurate the prediction is, and backs it with a firm offer to buy the vehicle.


WHAT WE DO

We predict what the damaged vehicle will fetch

A motor total loss in Germany or Italy is settled as the pre-accident value less the residual value, paid in cash, with the claimant keeping the vehicle. The residual value is therefore a deduction from the payout — and until now it has been whatever the available bidders happened to offer that week.

Replacement value — an equivalent undamaged car

This is the figure we predict.

The claimant keeps the damaged vehicle and sells it. Every euro of residual value is a euro the insurer does not pay — which is why this is the figure the whole settlement turns on.

Germany and Italy. The proportions are illustrative and differ from vehicle to vehicle.
THE PREDICTION

The highest bid the vehicle would draw

Across every residual-value platform and across borders, returned in seconds from the data already on the claim — with a confidence range around it, not a bare point estimate.

THE EVIDENCE

Accuracy we publish, including the weak parts

Measured on claims removed from the training data, and reported by segment so the parts where the model is least accurate are visible rather than averaged away.

THE COMMITMENT

A firm offer behind the figure

Where an insurer wants to act on the number, we commit to buy the vehicle at it, collect it and sell it ourselves — through our own settlement infrastructure, which is what makes the commitment a mechanism rather than a promise.

In Germany and Italy the claimant owns the damaged vehicle throughout and the insurer never takes title, so the settlement figure is the only thing in play. France works differently — there the insurer acquires the vehicle — which is why the same product does not transfer across that border unchanged.

THE EVIDENCE

We publish the error band. Nobody else in this market does.

Four things decide whether an accuracy figure means anything. All four are stated here, and all four are put in writing to an insurer evaluating the model.

Blind
TESTED ON CLAIMS IT HAS NEVER SEEN

Recent claims are withheld from training entirely, then scored. A model measured on the data it learned from is not being measured at all.

Realised
OUTCOMES, NOT ESTIMATES

Every training row is a vehicle that went through the process and produced a real market figure — not another expert's opinion of one.

By segment
INCLUDING WHERE IT IS WEAKEST

Accuracy is reported per segment as well as in total, so the weakest part is visible before a pilot rather than after it.

On your book
RECALIBRATED, THEN SCORED BLIND

The pilot re-fits on your earlier claims and scores your recent ones. The number that decides this is measured on your portfolio, not ours.

Accuracy is measured out-of-sample: the claims used to test the model are removed from the data used to train it, so the figure is not a self-assessment. We disclose the segments where the model is weakest as well as where it is strongest — a portfolio number that hides its own tail is not evidence. The measured figures themselves, and the figures we derive from an insurer's own book, are shared directly with insurers rather than published here.

HOW IT WORKS

Nothing in the claims process has to change

1

The claim file arrives

The assessment reaches you exactly as it does today. Nothing changes for the assessor or the claimant.

2

Scored in seconds

Through an API on the live claim, or in batch over a file, using the data already on the claim.

3

A value and a range

Not a single point estimate — a full distribution, so you can see how confident the model is on that vehicle.

4

A firm offer, if you want one

Where you want to act on the figure, we commit to buy the vehicle at it — and our own settlement process collects it, pays for it and handles the documents.

Delivery is not the obstacle: ECX has been integrated in the largest German assessor system since 2024, and insurer posting agreements are in place. Buyer economics are modelled separately for 14 markets rather than as one European average.

Send one file. See the difference on your own claims.

Twelve months of total losses, calibrated on the earlier months with the recent ones held back. You see our prediction against what you actually settled at, claim by claim.

Start there